Skip to content

Commit 44e489b

Browse files
committed
adjust shortlist description
1 parent 5264297 commit 44e489b

File tree

2 files changed

+2
-2
lines changed

2 files changed

+2
-2
lines changed

latex/popl2026/popl.pdf

-55 Bytes
Binary file not shown.

latex/popl2026/popl.tex

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -699,7 +699,7 @@ \subsection{Disambiguation}\label{sec:transfer_method}
699699
\item $\texttt{lev\_build}: \Sigma^n \times \mathbb{N} \rightarrow \text{NFA}$ -- constructs a Levenshtein NFA from the broken string.
700700
\item $\texttt{cfl\_fixpt}: \text{NFA} \times \text{CFG} \rightarrow \mathbb{B}^{|Q|\times |Q| \times |V|}$ -- computes the matrix exponential.
701701
\item $\texttt{reg\_build}: \mathbb{B}^{|Q|\times |Q| \times |V|} \times \text{CFG} \rightarrow \text{GRE}$ -- constructs the regular expression for $G_\cap$.
702-
\item $\texttt{reg\_dcode}: \text{GRE} \times \mathbb{N}^{|\Sigma|^{c\approx 4}} \hspace{-0.05cm}\times \mathbb{N} \rightarrow\hspace{-0.02cm} (\Sigma^+)^{k\approx 10}$ -- returns a small set of the most probable repairs.
702+
\item $\texttt{reg\_dcode}: \text{GRE} \times \mathbb{N}^{|\Sigma|^{c\approx 4}} \hspace{-0.05cm}\times \mathbb{N} \rightarrow\hspace{-0.02cm} (\Sigma^+)^{k\approx 10^4}$ -- returns a short list of high-probability repairs.
703703
\end{enumerate}
704704

705705
\noindent We will now explore the imperative pseudocode for each stage, starting with the Levenshtein automata constructor, which is a straightforward translation of the inference rules in \S~\ref{sec:repair_ex}.
@@ -795,7 +795,7 @@ \subsection{Disambiguation}\label{sec:transfer_method}
795795
\end{algorithmic}
796796
\end{algorithm}
797797

798-
\noindent Now, we have our shortlist of repairs and after cosmetic postprocessing, can present them to the user. With this approach, we can quickly generate a representative subset of $\ell_\cap$ within a fixed latency budget, e.g., 100ms, or otherwise terminate early should we succeed in exhaustively generating it.
798+
\noindent With this approach, we can quickly generate a representative subset of $\ell_\cap$ within a fixed latency budget, e.g., 100ms, or else terminate early should we succeed in exhaustively generating it. After reranking, truncation, and cosmetic postprocessing, the resulting repairs can then be displayed.
799799

800800
\clearpage\subsection{GPU translation}\label{sec:gpu_translation}
801801

0 commit comments

Comments
 (0)